Earlier quoted context omitted.
I think you greatly underestimate the programming challenges available in large investment banks, from "We ingest literally thousands of separate sources, many of them time-series, at the rate of gigabytes per second." to "How does one manage a programming team of ten thousand people split between six countries?" to "Where do we strike the balance between the-most-fault-tolerant-computer-is-one-that-is-powered-off an…
But these are only tangentially related to the real-world in the sense that a over-a-century-old monopolist family & corporation has constructed systems to enrich themselves and requires maintenance on those systems to continue to extract ever more currency from their customers (or non-customers, aka 'market', as it may be). Most people's conception of real-world problems would be more along the lines of making sure…
Morgan Stanley has plenty of "real problems" in that they have plenty of challenging and exciting technical problems that rival those you would see at a top tech company. It isn't fair to dismiss someone's reasons for joining a top Wall St firm as purely financial, because it has many technical challenges to offer very skilled engineers.
Couple this with the fact that, let's face it, most people at AmaGooBookSoft simply aren't "changing the world" in the moral revolutionist sense of the term, and that most of the top tech companies have so much technical debt that many of their engineering teams are focused on maintaining old CRUD apps.
I can perfectly understand Stroustrup's decision here. You'll almost always be sorely disappointed if you join a company for its glamorous reputation. On the other hand, it's objectively verifiable that some companies have truly challenging technical problems, and others don't.